The 5% Router Tax: What Hosted LLM Gateways Charge For (and How to Self-Host It)

说明与用途

自托管代理（Lynkr）可消除5%令牌费用，保持请求和数据本地化，并为大多数流量使用本地模型，仅将稀有模型保留给云路由器。

问题

Hosted LLM routers charge a ~5% fee per token and transit all prompts, code, and secrets, which can be costly and raise compliance concerns for coding workloads.

结果

The hybrid approach removes the 5% router fee for the majority of traffic, keeps prompts and code local, allows token optimization before billing, and retains access to new models via a hosted router for rare use cases.

工作流

• 1. Install Lynkr: npm install -g lynkr
• 2. Initialize: lynkr init
• 3. Start proxy: lynkr start
• 4. Configure routing rules: simple/medium → local Ollama/llama.cpp, complex → direct provider API keys, exotic → OpenRouter
• 5. Monitor and update proxy as needed

证据

The price is a ~5% fee on every token you route (5.5% on OpenRouter credits, 5% on Requesty)

来源

A self-hosted proxy can strip unusable tool schemas (measured: −53% on tool-heavy requests) and compress JSON tool results (measured: 3,458 → 427 tokens) before any provider bills you.

来源